stats Importance of sample-size and WR

[TheDawnSentinelRP]

Hey there!

Been wondering about this for a while now, my WR in the ships i play a lot seems to be all over the place atm.

 

So to get to the point why I am starting this topic:

Ive been kinda hovering between 54-55% account wr for a while now, thats pretty much where I see myself gameplay-wise as well, dont really own any of the OP-ships and dont usually sealclub. Most of the time I play T8 atm, probably my favorite tier tbh.

The last two weeks have been a rollercoaster wr-wise tho, daily wr ranging from 25-80%. That kind of has taken a toll on my individual ship winrates, as an example ill take my Akizuki.

I used to have a 58% wr (50 battles in), well that has plummeted to 52% yesterday, while my winrate in the Kitakaze has skyrocketed to 65%. I basically had a winstreak on Kitakaze while having a severe loosing-streak on Akizuki (last 8 matches played in her almost)

I was thinking to myself that doesnt make any sense. These two ships are very similar, kinda the same ship actually, so how can it be that my wr are so all over the place?

Same in my Harekaze, got quite unlucky when I got her, two big loosing streaks the first day I got her but managed to get back up to 53% (60 games in). I mean, I dont think wr is all luck but this is just weird and frustrating.

Ive noticed that on some ships it seems I cant loose, even if I went afk, on some the matchmaker is out to get me for some reason.

 

So back to the topic, how big does the sample size have to be to accurately represent the "skill" of a player? Ive noticed that win and loosing streaks of blowout-matches influence my wr sometimes up to 10% (in ships ive 100 matches in).

[ColonelPete]

Kitakaze is better than Akizuki, as she is more nimble.

 

With 100 battles, 90% of the players will be around +9/-9% points around their true winrate.

With 500 battles, 90% of the players will be around +4/-4% points around their true winrate.

With 5000 battles, 90% of the players will be around +1,5/-1,5% points around their true winrate.

 

That means your true Akizuki winrate could be higher than 61% or even lower than 43%. Your sample size is just too small.

 

 

[TheDawnSentinelRP]

7 minutes ago, ColonelPete said:

Kitakaze is better than Akizuki, as she is more nimble.

 

With 100 battles, 90% of the players will be around +9/-9% points around their true winrate.

With 500 battles, 90% of the players will be around +4/-4% points around their true winrate.

With 5000 battles, 90% of the players will be around +1,5/-1,5% points around their true winrate.

 

That means your true Akizuki winrate could be higher than 61% or even lower than 43%. Your sample size is just too small.

 

 

Is ship-wr then even viable as "skill-representation"? I mean its rare to play even 100 games in a single ship.

I actually like Akizuki more because of the ships you are normally facing when playing her (less mogadors pretty much :D)

[ColonelPete]

4 minutes ago, TheDawnSentinelRP said:

Is ship-wr then even viable as "skill-representation"? I mean its rare to play even 100 games in a single ship.

I actually like Akizuki more because of the ships you are normally facing when playing her (less mogadors pretty much :D)

Normally not, that is why you look at account solo WR.

 

But you can look at multiple ships of the same type and Tier and if there is a pattern (mostly good or bad WR), it can give you information about the skill of a player. But it is usually better to look at account WR.

[tocqueville8]

36 minutes ago, TheDawnSentinelRP said:

I mean its rare to play even 100 games in a single ship.

 

Seriously, though: if you're free-to-play a T9--->T10 grind can take 100 games, easily.

 

It's in the area of 300k xp when you include the upgrades for the T9 ship, and a decent player might do about 1.5k base xp, on average. Throw in a +50% flag and a +50% first-win-of-the-day bonus, or neither, but a +100% camo instead, and it takes roughly 100 games.

[TheDawnSentinelRP]

1 hour ago, tocqueville8 said:

 

Seriously, though: if you're free-to-play a T9--->T10 grind can take 100 games, easily.

 

It's in the area of 300k xp when you include the upgrades for the T9 ship, and a decent player might do about 1.5k base xp, on average. Throw in a +50% flag and a +50% first-win-of-the-day bonus, or neither, but a +100% camo instead, and it takes roughly 100 games.

I went in about 30 games from Aki to Kita, another 40 and im about 40k exp short of harugumo. I rarely play more than 50 games (as part of the grind).

[tocqueville8]

2 minutes ago, TheDawnSentinelRP said:

I went in about 30 games from Aki to Kita

Standard account? No xp flags, no camo? That's mighty impressive!

[TheDawnSentinelRP]

1 hour ago, tocqueville8 said:

Standard account? No xp flags, no camo? That's mighty impressive!

Nope, premium account (got lucky and pulled 24 days i think from a super-container) with economic flags (i got tons of them) and sometimes camos ive laying in storage. 

[RenamedUser_92906789]

4 hours ago, TheDawnSentinelRP said:

Nope, premium account (got lucky and pulled 24 days i think from a super-container) with economic flags (i got tons of them) and sometimes camos ive laying in storage. 

 

Here, If you have not claimed yet (make sure you login first).

 

https://worldofwarships.eu/en/news/general-news/14-days-of-premium-account/

[totally_potato]

8 hours ago, TheDawnSentinelRP said:

Hey there!

Been wondering about this for a while now, my WR in the ships i play a lot seems to be all over the place atm.

 

So to get to the point why I am starting this topic:

Ive been kinda hovering between 54-55% account wr for a while now, thats pretty much where I see myself gameplay-wise as well, dont really own any of the OP-ships and dont usually sealclub. Most of the time I play T8 atm, probably my favorite tier tbh.

The last two weeks have been a rollercoaster wr-wise tho, daily wr ranging from 25-80%. That kind of has taken a toll on my individual ship winrates, as an example ill take my Akizuki.

I used to have a 58% wr (50 battles in), well that has plummeted to 52% yesterday, while my winrate in the Kitakaze has skyrocketed to 65%. I basically had a winstreak on Kitakaze while having a severe loosing-streak on Akizuki (last 8 matches played in her almost)

I was thinking to myself that doesnt make any sense. These two ships are very similar, kinda the same ship actually, so how can it be that my wr are so all over the place?

Same in my Harekaze, got quite unlucky when I got her, two big loosing streaks the first day I got her but managed to get back up to 53% (60 games in). I mean, I dont think wr is all luck but this is just weird and frustrating.

Ive noticed that on some ships it seems I cant loose, even if I went afk, on some the matchmaker is out to get me for some reason.

 

So back to the topic, how big does the sample size have to be to accurately represent the "skill" of a player? Ive noticed that win and loosing streaks of blowout-matches influence my wr sometimes up to 10% (in ships ive 100 matches in).

WR doesn't mean that you are a bad player nowadays. Terrible teams lose games so damn much that WR is nothing more than paper value to me. I have seen many 60% WR players that do mistakes that the average potato player might have corrected by now, and I see sub 50% WR players, play like a super-unicum. WR is overrated and you should not care about it. My current WR is 49% and it 56% in randoms, before the commander rework. But I just don't care coz my skill hasn't changed, its the team who just do crap things on a regular basis. 

 

That's why I hate stat shaming from ppl like Flamu coz it really doesn't tell you anything. I have seen many players (from Flamu and Flambass's community) who claim that WR is very accurate, when it clearly misleads. 

[GarrusBrutus]

9 hours ago, TheDawnSentinelRP said:

how big does the sample size have to be to accurately represent the "skill" of a player?

After 1000 solo games you can accurately tell how good a player is.  People may have bad luck, back teams, play bad ships, etc. But after 1000 games those factors are of little influence. Ofcourse a player who plays 1000 Erie games, or a player who plays 1000 Enterprise games are very specialized in one ship and might not be very good at any other ship. But a player who plays all classes and different ships and still has 60% solo winrate after a 1000 games, is considered a very good player. 

[ReapingKnight]

1 hour ago, totally_potato said:

WR doesn't mean that you are a bad player nowadays. Terrible teams lose games so damn much that WR is nothing more than paper value to me. I have seen many 60% WR players that do mistakes that the average potato player might have corrected by now, and I see sub 50% WR players, play like a super-unicum. WR is overrated and you should not care about it. My current WR is 49% and it 56% in randoms, before the commander rework. But I just don't care coz my skill hasn't changed, its the team who just do crap things on a regular basis. 

 

That's why I hate stat shaming from ppl like Flamu coz it really doesn't tell you anything. I have seen many players (from Flamu and Flambass's community) who claim that WR is very accurate, when it clearly misleads. 

Whilst I agree on the stat shaming issue, I do believe there is more value in WR than you say.

 

Many forum posts have brought it up and from memory it boils down to if you as a player have played 1000 games and have a 55% WR chances are you are better than your avg potato as you've won more, in more specifics there will obviously be games where a team pulls you down but there are one's where the team will carry you. People say that these balance out so then the remaining games where the teams are evenly split or at least close will be determined by the one constant factor which would be your position on the team.

[ColonelPete]

2 hours ago, totally_potato said:

WR doesn't mean that you are a bad player nowadays.

That is what bad and average players say...

[totally_potato]

35 minutes ago, ColonelPete said:

That is what bad and average players say...

Like I said

WR is overrated

And I don't need you to say that I am bad or average

I will be the judge of that

[Taliesn]

3 hours ago, totally_potato said:

WR doesn't mean that you are a bad player nowadays. Terrible teams lose games so damn much that WR is nothing more than paper value to me. I have seen many 60% WR players that do mistakes that the average potato player might have corrected by now, and I see sub 50% WR players, play like a super-unicum. WR is overrated and you should not care about it. My current WR is 49% and it 56% in randoms, before the commander rework. But I just don't care coz my skill hasn't changed, its the team who just do crap things on a regular basis. 

 

That's why I hate stat shaming from ppl like Flamu coz it really doesn't tell you anything. I have seen many players (from Flamu and Flambass's community) who claim that WR is very accurate, when it clearly misleads. 

 

Your WR on any 5 game stretch doesn't mean anything. Your WR after thousands of battles does.

 

1 minute ago, totally_potato said:

Like I said

WR is overrated

And I don't need you to say that I am bad or average

I will be the judge of that

 

You are the only one who can judge that because you have hidden stats. Otherwise, nope.

 

I've yet to see a good player claim WR is overrated.

[DFens_666]

2 hours ago, totally_potato said:

WR doesn't mean that you are a bad player nowadays. Terrible teams lose games so damn much that WR is nothing more than paper value to me. I have seen many 60% WR players that do mistakes that the average potato player might have corrected by now, and I see sub 50% WR players, play like a super-unicum. WR is overrated and you should not care about it. My current WR is 49% and it 56% in randoms, before the commander rework. But I just don't care coz my skill hasn't changed, its the team who just do crap things on a regular basis. 

 

That's why I hate stat shaming from ppl like Flamu coz it really doesn't tell you anything. I have seen many players (from Flamu and Flambass's community) who claim that WR is very accurate, when it clearly misleads. 

 

Here is why this doesnt make sense. So you say, that WR is irrelevant, and people are only being lucky or unlucky with their teams (which already is questionable, if one would be lucky/unlucky ALL the time). But lets say, someone is lucky and gets the "winning" team all the time, thus he has better players on his team more frequently. So they win more often, right? But you said, WR is irrelevant and doesnt represent skill, so how can that be? So you claim, WR would be irrelevant for a small number of players, who are just lucky all the time, but the rest of his teammates, who are carrying him are actually good players and their WR is accurate.

This is just too far fetched in my book.

 

If you see 60% players play like dogpoo, you have to dig deeper into their stats to see whats going on. Maybe they are T2 mains, who occasionally play hightiers with 40% WR. Or maybe they are division players who are otherwise not so great. I had one of those, who had 60% WR on his acc. When he played solo, it was only 50%. And when he played hightier, it was <45%. So he was just pushing his stats with division and low/midtier solo games, but when he was in TX ranked, he was the usual red-potato who didnt know what to do.

[ColonelPete]

8 minutes ago, totally_potato said:

Like I said

WR is overrated

And I don't need you to say that I am bad or average

I will be the judge of that

No. Your stats say that.

 

[totally_potato]

3 minutes ago, DFens_666 said:

 

Here is why this doesnt make sense. So you say, that WR is irrelevant, and people are only being lucky or unlucky with their teams (which already is questionable, if one would be lucky/unlucky ALL the time). But lets say, someone is lucky and gets the "winning" team all the time, thus he has better players on his team more frequently. So they win more often, right? But you said, WR is irrelevant and doesnt represent skill, so how can that be? So you claim, WR would be irrelevant for a small number of players, who are just lucky all the time, but the rest of his teammates, who are carrying him are actually good players and their WR is accurate.

This is just too far fetched in my book.

 

If you see 60% players play like dogpoo, you have to dig deeper into their stats to see whats going on. Maybe they are T2 mains, who occasionally play hightiers with 40% WR. Or maybe they are division players who are otherwise not so great. I had one of those, who had 60% WR on his acc. When he played solo, it was only 50%. And when he played hightier, it was <45%. So he was just pushing his stats with division and low/midtier solo games, but when he was in TX ranked, he was the usual red-potato who didnt know what to do.

No that's not what I meant to say

Its unimportant to me not irrelevant. 

I am saying that individually stats don't paint the whole picture, and what you show is what you are judged upon, which is why the whole idea of stat shaming is trash. 

All I am saying that WR can be misleading unless its so bad like 30% or 25% or so good like 70%.

That's all I am saying here

No hate

[totally_potato]

2 minutes ago, ColonelPete said:

No. Your stats say that.

 

My stats only show my performance based on my wins, which have a lot contributing to it than just pure skill which is the whole point

 

[ColonelPete]

22 minutes ago, totally_potato said:

My stats only show my performance based on my wins, which have a lot contributing to it than just pure skill which is the whole point

 

Not with 1000th of battles. That is the whole point.

[GarrusBrutus]

29 minutes ago, totally_potato said:

My stats only show my performance based on my wins, which have a lot contributing to it than just pure skill which is the whole point

 

Such as? 

[GraySlayer]

Everytime this WR topic comes up im gonna treat myself to a small cake and get fat..

[bobtherterrible]

3 hours ago, GarrusBrutus said:

After 1000 solo games you can accurately tell how good a player is.  People may have bad luck, back teams, play bad ships, etc. But after 1000 games those factors are of little influence. Ofcourse a player who plays 1000 Erie games, or a player who plays 1000 Enterprise games are very specialized in one ship and might not be very good at any other ship. But a player who plays all classes and different ships and still has 60% solo winrate after a 1000 games, is considered a very good player. 

So the thing with WR and averages is that if you play a lot of games while you're Terrible^TM it takes a lot of better games to fix your average.

Personally I'm enjoying the challenge of not losing and it does help to reflect on if you are making the right choices to get a win each time.

 

Also, the correct sample size is 21

No particular reason for this magic number.......

 

 

[Bear__Necessities]

You can tell by the consistency of a player over 1000's of games. WR and other factors after 1000 games such as Kill Death and to an extent PR (though PR can be HEAVILY skewed) show a players worth. Of of the things I've found is between my three accounts on EU and NA. That I'm consistent, even taking k to account the 1000's of games played when I was new and a full blown potato. 

 

I'm happy to use mine as an example. The 1st (EU) and 2nd (main NA) show that while my EU doesn't look as impressive, my EU also factors all my learning games. But it does match a lot of the same numbers as my more recent NA account, showing the consistency. The 3rd is a an almost pure solo account I made as a personal challenge. That also highlights how stats can be interpreted with or without context.

 

EU

 

NA 1

 

NA 2

 

[totally_potato]

3 hours ago, GarrusBrutus said:

Such as? 

your team, the enemy team, the ship you are playing in that match, enemy team's counterplay or lack of counterplay, and luck. 

There is a saying that good players make their own luck, but at the end of the day its luck

Speaking of luck, RNG also contributes to your WR.

So my point here is WR doesn't tell you the whole picture.